<p>Welcome to <b>OmicSelector</b> - the software intended to find the best biomarker signiture based on NGS (miRNA-seq, RNA-seq) and
<p>Welcome to <b>OmicSelector</b> - the software for finding the best biomarker signiture based on NGS (miRNA-seq, RNA-seq) and
qPCR data.</p>
<divclass="panel-group">
<?phpif($_GET["msg"]!=""){?>
...
...
@@ -122,10 +122,10 @@
<p><b>How to prepare files for the analysis?</b>
<ul>
<li>All variables which name starts with <code>hsa</code> will be considered a features of interest. Features of interest must be numeric and has no missing values.</li>
<li>All variables which name doesn't start with <code>hsa</code> will be considered a metadata.</li>
<li>The dataset has to have the variable <code>Class</code> with no missing data and with values expicitly encoded as <code>Cancer</code>or<code>Control</code> cases.</li>
<li>The pipeline will perform data spltting in ratio 60% (training set) : 20% (testing set) : 20% (validation set). If you wish to enforce your way of spltting, the submitted file should have the variable named <code>mix</code> with values <code>train</code>, <code>test</code> and <code>valid</code>.</li>
<li>All variables which names start with <code>hsa</code> will be considered as features of interest - they must be numeric and have no missing values.</li>
<li>All variables which names don't start with <code>hsa</code> will be considered as metadata.</li>
<li>The dataset must contain the <code>Class</code>variable with no missing data and with values expicitly encoded as <code>Cancer</code>and<code>Control</code> cases.</li>
<li>The pipeline will perform data spltting in 60% (training set) : 20% (testing set) : 20% (validation set) ratio. If you wish to import your own splitting, the data file should have the variable named <code>mix</code> with values <code>train</code>, <code>test</code> and <code>valid</code>.</li>
</ul>
</p>
<p><ahref="https://github.com/kstawiski/OmicSelector/blob/master/example/Elias2017.csv"target="_blank">See exemplary file: <code>Elias2017.csv</code></a>. This file originates from our paper <ahref="https://elifesciences.org/articles/28932"target="_blank">Elias et al. 2017</a>.
